Improving business performance and customer satisfaction with free text mining
The challenge
How does an organisation develop internal and external strategies to improve business performance and increase overall customer satisfaction - especially when it is difficult to quantify the underlying issues facing different customer groups in the first place?
ACC’s business service centre logs all customer-related interactions in a system called Talisman - including inbound, outbound and internal communications. While comprehensive in the amount of information Talisman holds, the system is essentially a free text record of all activity within the business service centre. Although notes are time stamped and linked to an ACC number, there was no other classification system for storing basic interaction activity such as whether a call is inbound, outbound or internal.
As part of ACC’s overall strategy of improving customer satisfaction, Datamine was asked to assist in analysing the wealth of information available. The main purpose of the project was to identify and understand groups of customers and the issues they face when dealing with ACC via the business service centre.
The solution
The specific objectives of the project were to:
- Determine the purpose of a call to the service centre in relation to an invoice eg. required more information, clarification, reaction to an invoice itself, or a debt related query
- Understand the profiles of different customer groups based on their amount and type of interactions with the business service centre
The objectives of this project were realised by incorporating and time-aligning call centre note data alongside other ACC data sources such as existing customer information and invoicing data.
Analysing the approximately 500,000 free text notes was only made possible by building an extensive data dictionary of words, as well as determining as many traditional and non-traditional abbreviations and permutations of those words as possible. From there the words within notes were looked at in context to each other, enabling Datamine to accurately assign over 90% of the notes to various classification levels such as call type (inbound, outbound, or internal), and even down to the exact reason for the note itself eg. a debt related query.
The results
This provided ACC with a comprehensive look at the profiles of various groups of customers, and the reasons for their interaction with the ‘business service centre’ at different levels.
The analysis enabled ACC to identify and change processes to better meet the needs of their customers - thereby increasing their level of satisfaction. These included simplifying collateral and creating new communications to meet specific needs.